Designing a Predictive Model for Train Arrival Time Management, Using Data Mining Approach

Designing a Predictive Model for Train Arrival Time Management, Using a Data Mining Approach

ABSTRACT

Accessible and efficient transportation services are paramount for the holistic development of a nation. They play a pivotal role in managing various facets of train operations, encompassing factors such as staff conduct, affordability, ticketing systems, reliability, passenger comfort, safety, security, accessibility, and availability. Nevertheless, the realm of transportation services is not devoid of challenges, with passenger loading being a primary concern for railway service providers. This research endeavors to create a predictive model tailored to ascertain the Train Arrival Time Management of the Addis Ababa Light Transit (AALRT) operating control center, utilizing its operational data. In response to the limitations of conventional statistical methods, we propose the application of data mining techniques to analyze data pertaining to train arrival time management. This study adheres to a hybrid data mining process model. Following an exploratory survey to comprehend the problem space, approximately 20,000 records spanning three years were extracted from the AALRT operating control center's dataset. Subsequently, through the elimination of irrelevant and superfluous data, a refined dataset consisting of 15,040 entries, each characterized by 12 attributes, was employed for the investigation. Rigorous data preprocessing procedures were executed to ensure data cleanliness and consistency. The refined dataset was then meticulously formatted in ARFF (Attribute-Relation File Format) to suit the data mining tasks. The research was executed utilizing WEKA software version 3.8 and employed three classification techniques: the J48 algorithm derived from decision trees, Naïve Bayes, and JRIP algorithm derived from rule induction. Notably, the J48 decision tree algorithm, with a data split percentage of 66%, exhibited superior performance, boasting an accuracy rate of 95.56%. The outcomes of this study shed light on the critical determinants of train arrival time management within the AALRT system. Factors such as speed, headway time, and passenger loading were identified as pivotal attributes significantly influencing the success of train arrival time management. Additionally, the research uncovered that train arrival time management in other regions exhibited a strong association with its success rate, underlining the broader implications of the findings beyond the confines of the AALRT system.

Designing a Predictive Model for Train Arrival Time Management, Using a Data Mining Approach, GET MORE COMPUTER SCIENCE PROJECT TOPICS AND MATERIALS

Sharing is caring!